AI Discovers An Unknown Human Ancestor

  • Researchers applied a new method in the genome of modern Asian people and found a third, previously unknown, species. 
  • This new method utilizes deep learning in an Approximate Bayesian Computation framework to uncover the mysteries of human evolution in ancient DNA. 

Us Humans are genetically connected to each other at a time depth of about 350 thousand years ago, and share a common African root. It’s well-known that modern humans carry fragments of DNA from two different extinct species: Neanderthals and Denisovans.

Recently, an international team of researchers developed a new method to statistically compare complex demographic models and evaluate demographic factors that can be applied to the evolution of introgressions in Europe and Asia.

Researchers employed this method in the genome of modern Asian people and discovered a third type of species that interbred with ancient human races. Although this third introgression has long been suspected in all Oceanian and Asian populations from an archaic population, this is the first time someone has explained this with the help of technology.

New Method For Explaining Complex Human Evolution

We all know that a small fraction of ancient humans abandoned the African continent and migrated to other continents approximately 80,000 years ago from now. This gave rise to another population that cross bred with Denisovans in Oceania and with Neanderthals (excluding Africa). However, we still have no solid proof of cross-breeding with a third extinct ancestor.

The existence of third extinct species remains a theory that explains some portions of the modern human genome. Now with the help technology, we can transform data obtained from DNA into the demographics of ancestral populations.

This new method utilizes the nonlinear capabilities of deep learning for detecting patterns and extracting features. It also leverages the Approximate Bayesian Computation framework for precisely evaluating the posterior distribution of parameter and models by generating simulated datasets.

Reference: Nature Communications | doi:10.1038/s41467-018-08089-7 | Center for Genomic Regulation 

This technique overcomes several limitations for evaluating different demographic models while minimizing manual fitting. Basically, researchers have applied deep learning algorithms on genomes extracted from hundreds of thousands of simulations. These algorithms can efficiently learn and predict human demographics, providing us a comprehensive view that makes the hereditary puzzle fit together.

third, unknown human ancestorA modern human and a Neanderthal skull | Image credit: hairymuseummat/DrMikeBaxter

The new approach supports a human evolution model in which Out-of-Africa population have had introgression from Denisovans and Neanderthals, as well as from the third species that is still archaeologically and genetically undefined.

What’s Next?

In this study, researchers have considered the multidimensional absolute site frequency spectrum as raw summary statistics, which accounts for the number of nucleotide polymorphisms present in a specific combination of sampled populations.

It doesn’t consider additional information, for example, nature and length of the introgressed fragments, which could be a crucial factor for discriminating between different models.

Read: “I-motif” – A Completely New DNA Structure In Human Cells

The current approach can be further improved by incorporating the fragment length of ancestors of the genomic fragments. Also, additional ancient genome sequences and new investigations are necessary to gain more insights into the nature and consequences of this third species in modern humans’ genome.

Written by
Varun Kumar

I am a professional technology and business research analyst with more than a decade of experience in the field. My main areas of expertise include software technologies, business strategies, competitive analysis, and staying up-to-date with market trends.

I hold a Master's degree in computer science from GGSIPU University. If you'd like to learn more about my latest projects and insights, please don't hesitate to reach out to me via email at [email protected].

View all articles
Leave a reply